Conference PaperPDF Available

Truck parking in urban areas: Application of choice modelling within traffic microsimulation

Authors:

Abstract and Figures

Urban truck parking policies include time restrictions, pricing policies, space management and enforcement. This paper develops a method for investigating the potential impact of truck parking policy in urban areas. An econometric parking choice model is developed that accounts for parking type and location. A traffic simulation module is developed that incorporates the parking choice model to select suitable parking facilities/locations. The models are demonstrated to evaluate the impact of dedicating on-street parking in a busy street system in the Toronto CBD. The results of the study show lower mean searching time for freight vehicles when some streets are reserved for freight parking, accompanied by higher search and walking times for passenger vehicles.
Content may be subject to copyright.
Nourinejad, Wenneman, Habib, & Roorda 1
Truck Parking In Urban Areas: Application of Choice Modelling Within Traffic 1
Microsimulation 2
3
4
Mehdi Nourinejad 5
Department of Civil Engineering, University of Toronto 6
35 St. George Street, Toronto, ON M5S 1A4 7
Telephone: +1 647 262 6234 8
Email: mehdi.nourinejad@mail.utoronto.ca 9
10
Adam Wenneman 11
Department of Civil Engineering, University of Toronto 12
35 St. George Street, Toronto, ON M5S 1A4 13
Telephone: +1 416 708 0667 14
Email: adam.wenneman@mail.utoronto.ca 15
Corresponding Author 16
17
Professor Khandker Nurul Habib 18
Department of Civil Engineering, University of Toronto 19
35 St. George Street, Toronto, ON M5S 1A4 20
Telephone: +1 416 946 8027 21
Email: khandker.nurulhabib@utoronto.ca 22
23
Professor Matthew J. Roorda 24
Department of Civil Engineering, University of Toronto 25
35 St. George Street, Toronto, ON M5S 1A4 26
Telephone: +1 416 978 5976 27
Email: roordam@ecf.utoronto.ca 28
29
30
31
32
Number of words: 6214 33
Number of figures and tables: 5 34
Total: 6147+5*250=7397 35
36
37
38
39
40
41
42
43
44
45
46
47
48
Nourinejad, Wenneman, Habib, & Roorda 2
ABSTRACT 1
Urban truck parking policies include time restrictions, pricing policies, space management and 2
enforcement. This paper develops a method for investigating the potential impact of truck parking 3
policy in urban areas. An econometric parking choice model is developed that accounts for parking type 4
and location. A traffic simulation module is developed that incorporates the parking choice model to 5
select suitable parking facilities/locations. The models are demonstrated to evaluate the impact of 6
dedicating on-street parking in a busy street system in the Toronto CBD. The results of the study show 7
lower mean searching time for freight vehicles when some streets are reserved for freight parking, 8
accompanied by higher search and walking times for passenger vehicles. 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
Nourinejad, Wenneman, Habib, & Roorda 3
1 INTRODUCTION 1
Central business districts (CBDs) are major destinations for goods pickup and delivery in urban centres. 2
“Last mile” delays in CBDs are one of the most expensive components of urban freight (1). In this “last 3
mile”, truckers must navigate congested urban streets and search for appropriate parking. When 4
parking is unavailable or inappropriately located, delivery vehicles frequently park illegally, often 5
considering the parking tickets as a cost of doing business. This cost is increasing over time. From 6
2006, to 2009 parking fines in Toronto increased 70%, and there is little evidence that illegal parking 7
problems are being reduced. In Toronto, FedEx, UPS and Purolator paid an estimated $2.5M in parking 8
fines in 2009 (2). 9
The problem is significant and growing. The Toronto CBD, for example, receives a daily 10
average of 81,000 packages from express delivery alone (2). Parking and loading spaces are limited in 11
the CBD because many buildings were constructed before the invention of the automobile. Increasing 12
land values have resulted in the conversion of surface parking lots to high-rise buildings, which in turn 13
are increasing the demands for goods delivery. 14
Freight parking issues are common in other North American cities as well. The U.S. 15
Department of Transportation (USDOT) together with the Federal Highway Administration (FHWA) 16
and the Office of Freight Management and Operations prepared a series of case studies documenting 17
best practices for urban goods movement. Reports were prepared for Washington D.C., Orlando, New 18
York City, and Los Angeles. The purpose of these studies is to investigate initiatives aimed at 19
mitigating congestion and improving efficiency of commercial vehicle operations (3). 20
Urban policy makers are in need of data and decision support tools to identify impacts of 21
parking policy scenarios such as dedicated on-street parking for commercial vehicles, time restrictions, 22
and pricing policy. Traffic simulation tools are increasingly popular for urban traffic analysis, however, 23
they do not currently provide sufficient representation of parking. Parking simulation models have been 24
developed, but these models are for passenger parking, which is behaviourally different than truck 25
parking. Econometric models of parking choices have also been developed, but again are limited to 26
passenger cars (4). 27
This paper explores the potential of truck parking policies and develops a novel tool for 28
assessing the impacts of parking policy. In Section 2, we provide a review of strategies for dealing with 29
truck parking. In Section 3, we develop a truck parking selection model using data from a truck parking 30
survey conducted in the summer of 2010. In Section 4, we develop a traffic simulation model for a 31
small study area in the Toronto CBD. The model specifically represents on-street parking, off-street 32
surface parking lots, parking garages, truck loading docks, and alleyways suitable for truck 33
loading/unloading. A binary logit model for parking selection is incorporated in this simulation 34
environment that is capable of assessing the traffic impact of changes in parking policy on truck parking 35
choice. The model is applied to test the impact of two simple truck parking scenarios on measures of 36
effectiveness such as time to find parking, walking distance to the final destination, and total network 37
travel time. In Section 5, we present the conclusions of this research and suggest future research 38
directions. 39
40
2 LITERATURE REVIEW 41
The literature can be usefully divided into three parts: freight parking policies, parking simulation 42
models, and parking choice models. In Section 2.1, we describe policies targeting freight vehicle 43
operations and in Sections 2.2 and 2.3 we review the literature on simulation methods and choice 44
models of parking policy analysis, respectively. 45
46
2.1 Freight Vehicle Parking Policies 47
In dense CBDs, curb space is a scarce resource with high demands from a variety of users. Curb space 48
management policies impact road congestion, business vitality, urban aesthetics, and pedestrian safety 49
Nourinejad, Wenneman, Habib, & Roorda 4
and comfort (5). On-street parking is often the focus of parking management practices where there is 1
not ample supply to fulfill the demand. Policy makers have generally responded to this problem by 2
promoting parking turnover using control time limits and parking pricing. Higher meter rates, on the 3
other hand, are endorsed by those who believe time limitations are challenging to monitor and enforce. 4
Shoup argues that parking meters can create curb vacancies by directing a portion of drivers to off-street 5
parking facilities (9). This would reduce cruising for curb parking which can reduce congestion. 6
Nevertheless, the generated revenue from parking meters can be spent on public improvement in the 7
metered neighborhoods. Pasadena is an example where charging market prices (off-street prices) for 8
curb parking has improved congestion and made the area safer and cleaner from the generated parking 9
revenue (6). Clearly, implementation of any passenger vehicle parking policy indirectly affects the 10
operations of freight vehicle deliveries even if they are not the targeted. Any policy that produces more 11
vacant spaces on the curbsides creates better opportunities for on-street freight parking. 12
In addition to the indirect effect of passenger vehicle parking policies on freight vehicles, 13
loading zone regulations and freight restrictions directly impact freight deliveries. In response to recent 14
freight vehicle operations issues, the Federal Highway Administration developed case studies in some 15
of the major cities of the United States (Los Angeles, New York City, Washington DC, and Orlando) to 16
document prominent goods movement strategies (3). Freight parking strategies employed in these cities 17
included time restrictions, pricing strategies, parking space management, and parking enforcement. 18
19
Time Restrictions 20
A common freight parking strategy used in many cities is time of day loading zone restrictions. The 21
goal of such time restrictions is to separate commercial vehicles and passenger vehicles in urban areas 22
temporally instead of spatially. In Manhattan, the New York City Department of Transportation 23
(NYCDOT) is planning to implement delivery windows to designate curbside parking for freight 24
vehicles in the morning and create better parking opportunities for passenger vehicles later in the day. 25
They have learned that 65% of all deliveries occur before 12 PM and granting exclusive parking access 26
to freight vehicles during these hours can reduce traffic congestion. A similar strategy is used in 27
Philadelphia where loading zone restrictions (subject to parking enforcement) encourage local 28
businesses to receive any deliveries before 10:00 AM (5). In (7), Jaller et al. estimated that, in 29
Manhattan, shifting approximately 20% of freight traffic to off-peak hours would minimize the number 30
of over capacity parking locations. 31
32
Pricing Strategies 33
Pricing strategies, in general, can encourage greater turnover of both passenger and freight vehicles to 34
create better parking opportunities for newly arriving vehicles. The District Department of 35
Transportation (DDOT) in Washington, DC has installed loading zone meters along K Street in 36
response to all-day parking of commercial vehicles. The meters charge commercial vehicles $1 per hour 37
and allow a limit of two hours for parking. The NYCDOT has also implemented a pricing strategy using 38
the Muni-meter program that uses an escalating rate structure of $2 for one hour, $5 for two hours, and 39
$9 for three hours. This strategy has led to considerable reductions of dwell times (160 minutes to 45 40
minutes) and highlights the impact of different hourly pricing combinations. 41
42
Space Management 43
Commercial vehicle operations efficiency can improve if ample curbside space is reserved for them. 44
The NYCDOT encourages smaller jurisdictions to designate part of the curbside or even individual 45
spaces to commercial vehicles. The DDOT and Downtown DC Business Improvement District (BID) in 46
Washington, DC have also extended loading zones from 40 feet to 100 feet in length on K Street and 47
moved commercial loading zones to the approach end of each block wherever possible. 48
49
Nourinejad, Wenneman, Habib, & Roorda 5
Parking Enforcement 1
Parking enforcement responds to lack of regard for parking regulations. For example, the Los Angeles 2
Department of Transportation (LADOT) has initiated an enhanced parking enforcement program called 3
“Tiger Teams”. The program deploys fifteen uniformed traffic control officials and ten tow trucks to 4
enforce parking violations during peak hours. Washington DC has also adopted a similar program of 5
parking enforcement on K Street in addition to its other curb-space management policies. The 6
NYCDOT reports that enforcement is a critical component for a successful curbside management 7
program. They implemented a pilot program incorporating enforcement in 2002 called THRU Streets 8
(8). This program consisted of the designation of THRU streets, where traffic flow was prioritized, and 9
non-THRU streets, where accessibility was prioritized. On THRU streets, parking was made available 10
on one side only. Enforcement was increased on THRU streets, with the goal of reducing illegal parking 11
and increasing curb clear time. On non-THRU streets, multi-space MUNI meters were installed on both 12
sides of the street, creating approximately 150 additional freight parking spaces in the study area. This 13
pilot program resulted in a decrease in travel times, an increase in network capacity, and increased the 14
percentage of streets free of illegally parked vehicles. 15
16
2.2 Parking Simulation 17
Parking equilibrium models, which are common in the literature (9, 10, 11, 12, 13), are formulated to 18
capture the relationship between various parking activity components such as price of parking and 19
distance to final destination. One of the major drawbacks in such models is lack of regard for the 20
dynamic nature of parking behavior. These models neglect walking time (distance) and parking 21
availability at different hours of the day. Walking distance (from parking spot to destination) is critical 22
for those highly sensitive to time (such as truck drivers making deliveries). Similarly, representation of 23
the supply of off-street parking can be crucial. Furthermore, the equilibrium models have crude 24
presentation of cruising time which is highlighted as one of the most important features of parking 25
behaviour in (14). 26
In an attempt to fill in this gap in parking research, Benenson et al. in (15) propose an explicitly 27
space sensitive dynamic model called PARKAGENT to simulate behaviour of individual drivers. This 28
model is structured for two groups of agents (residential and visitor) in the city of Tel Aviv where future 29
surface parking construction is expected. In this research drivers enter the simulation when within 250 30
m (820 ft) vicinity of their destination and lower their speeds to 25 km/h where they become aware of 31
the need to park. This model is structured to evaluate the impact of additional parking facilities in the 32
residential area but fails to capture the impact of the type of parking facility. 33
Dieussaert et al. in (16) introduce SUSTAPARK which is a spatio-temporal tool simulating 34
both traffic and parking behavior in a cellular automata structured network. Their parking behaviour 35
model uses an MNL model which is a simplified version of a mixed logit model proposed by Hess et al. 36
in (17). In this MNL structure, agents select their initial parking type from two possible choices (on-37
street and off-street parking) based on parking features such as search time, egress time, and expected 38
fee which are input to a utility function. However, this initial choice is modified every 30 seconds 39
according to the re-evaluated parameter values and new utility function. The utility function value for 40
every agent would decrease as cruising time builds up to the point where parking off-street is a more 41
suitable option. 42
In (18) Munuzuri et al. develop a dynamic parking model for freight vehicles in a microscopic 43
traffic simulator called TRAMOS. In this model private and freight vehicles are assigned different 44
available types of parking facilities, parking choice behaviour, and number of stops for each vehicle 45
type. Private vehicles are assigned one stop whereas freight vehicles are assigned an itinerary of stops, 46
each of which has a specified location and duration. The freight parking choice model is based on a 47
weighting system for each parking facility type as a function of distance to delivery point. For example, 48
at up to 15 metres from the delivery point, the choice of loading zone parking has a higher decision 49
Nourinejad, Wenneman, Habib, & Roorda 6
weight than on street parking and is more likely to be chosen. The parking choice model for private 1
vehicles, however, is simpler and only function of the distance covered from the vehicle’s origin. 2
Private vehicles are assumed to take the first parking facility when the distance they have travelled has 3
reached a maximum threshold which is 1.25 times the width of the network. This proposed model is 4
only tested on a simple network of four nodes and three links and sufficient analysis is not provided to 5
assess policy. 6
Waraich and Axhausen in (19) extend MATSim to capture the influence of parking on daily 7
(activity) plan features including travel time, travel mode, and destination choice. Their model reflects, 8
for example, that insufficient or expensive parking may encourage drivers to change their mode of 9
transport or time of departure. MATSim agents iteratively select, among possible daily plans with 10
different utility scores, the one with the best final score (the “fittest alternative”). One of the 11
disadvantages of the proposed model by Waraich and Axhausen in (19) is that it is not completely 12
dynamic, meaning that perfect knowledge of parking availability is available before the trip is made. 13
Clearly, this overlooks the possibility of cruising for parking in cases where drivers arrive to their 14
destination and then start looking for a spot. 15
16
2.3 Parking Choice Modelling 17
Since (20), only (18) has considered commercial vehicle parking choice. Axhausen and Polak in (21) 18
estimated two parking choice logit models, one in Germany and the other in the UK. The data used in 19
their research was collected using a stated preference (SP) survey, which the researchers argued was 20
superior to previous studies using revealed preference (RP) data. They found that access time, parking 21
search time, walking time between the parking location and final destination, parking type, and parking 22
fee were all significant factors in selection of a parking location. Another important result from this 23
work was that time spent searching for parking and time spent driving to the general location had 24
significantly different parameters in Germany and the UK. 25
Teknomo and Hokao in (22) applied a multinomial logit model in Indonesia using an RP data 26
set. Similar to other studies (20, 21), they found that walking time, trip duration, parking fee, and 27
parking search time were all significant factors in selecting a parking location. Teknomo and Hokao in 28
(22) also found that the parking location choice depends on trip purpose, further supporting the findings 29
reported in (21). 30
Thompson and Richardson in (23) present a parking choice model based on expected gain in 31
utility. The main difference with this model was that unlike others that assumed the set of all parking 32
choices was known, their model allowed for vehicles to compare the utility of parking in the current 33
location against the likelihood that a better spot was available between their current location and the 34
destination, without having information about the area in between. 35
36
3 DATA AND METHOD 37
The new data collected for this research included a survey of truck drivers, a count of truck parking 38
events and a complete inventory of parking supply in the Toronto CBD (area between Queen St., 39
Simcoe St., Front St. and Victoria St). This inventory consists of on-street parking, alleyways, alleyway 40
loading zones, loadings bays, surface parking lots, and off-street public parking garages (Fig. 1). In 41
August of 2010, driver interviews and truck parking counts were conducted to determine the demand 42
for parking and loading. The interviews of truck drivers were conducted by a surveyor who targeted 43
parked commercial vehicles on individual road segments on weekdays between the hours of 9:00 AM to 44
3:00 PM. The interviews collected arrival time, departure time, parking location, type of vehicle, the 45
company that owned the commercial vehicle, the commodity delivered and the final destination of the 46
delivery. The survey instrument can be found in (24). While conducting surveys, the interviewer also 47
counted the total number of trucks parking in the road segment. Overall, 200 driver interviews and 48
observation of 1940 parking events were conducted. On average, approximately 10% of trucks parking 49
Nourinejad, Wenneman, Habib, & Roorda 7
in each segment were subject to a driver interview. A broad variety of commercial vehicle types and 1
commodity types were covered in the survey, resulting in a reasonable representation of truck 2
movements across the Toronto CBD. Details of the data collection effort are presented in (24). 3
Figure 1 shows the area in the Toronto CBD that was selected as the study area. The locations 4
marked with white squares on this figure represent some of the 60 most heavily ticketed locations in 5
Toronto as reported by the Canadian Courier Logistics Association (CCLA). The locations marked with 6
black squares are among the 10 most ticketed locations. This area also contains a mix of major two-way 7
arterial streets (Bay, Queen, and Yonge), major one-way streets (Richmond and Adelaide), and small 8
backstreets (York, Temperance, and Sheppard). The area consists mostly of high rise buildings 9
including the Bay Adelaide Centre (51 storey office complex), the Sheraton Centre (43 storey hotel), 10
the Richmond Adelaide Centre (12 storey office complex) and several other 12 20 storey office 11
towers. Retail and dining establishments are present at street level and office space is generally located 12
above street level. 13
14
15
FIGURE 1 Study area in the Toronto CBD. 16
17
The modelling methods developed in this paper include a parking choice model and a parking 18
simulation model. These models are described in the following sections. 19
Nourinejad, Wenneman, Habib, & Roorda 8
3.1 Parking Choice Model 1
The parking choice model is an econometric discrete choice model of parking spot selection. A binary 2
logit model is developed to determine the probability (!
!) of parking at a location. The alternative is to 3
reject the location in the hope of finding a better parking spot. This model can be written as (25): 4
!!=!!!!
!!!!!! (1) 5
where ! is a vector of estimated parameters and !! is a vector of characteristics of the current parking 6
location i. The binary logit model is estimated with data from the driver interviews, in which the 7
selected parking location was identified. The data were processed to identify the last two parking 8
locations that driver would have passed and rejected en route to his chosen parking location, as follows. 9
First, the address of the parking event and the address of the previous stop were found. Next, Google 10
Maps was used to find the driving route from the previous stop to the parking event. From the parking 11
inventory, the previous two appropriate parking facilities (i.e able to accommodate the vehicle type) that 12
the driver would have passed en route to the parking location were identified (if such facilities existed). 13
Finally, the walking distance to the delivery destination and other relevant attributes of the parking spot 14
were determined. 15
The binary logit model for freight vehicle parking location choice is sensitive to parking 16
availability, distance from the final shipment destination, and parking facility type. The parameters of 17
this model were estimated using maximum likelihood estimation. The estimated values for these 18
coefficients are statistically significant if the absolute value of the ‘t’ statistic is greater than 1.96 for the 19
95% confidence interval. The estimated parameters are summarized in Table 1. 20
The final model achieved a pseudo-R squared value of 0.3086. The negative coefficient on the 21
distance term shows that the further a parking space is from the delivery destination, the less likely it is 22
that a truck driver will choose to park there. The negative coefficient on the term representing on street 23
parking reveals a preference against parking on street. Conversely, the positive coefficient on the term 24
representing loading bays is positive, revealing a preference towards parking in loading bays. Other 25
parking facilities did not enter the model as their coefficients were not statistically significant. 26
27
TABLE 1 Binary Choice Model for Freight Vehicle Parking Location
Log Likelihood
-84.35
Pseudo R-squared
0.3086
Variable
Coefficient
t-stat
Distance to destination
-6.23E-03
-3.87
On street parking facility
-1.61
-4.11
Loading bay parking facility
2.21
2.09
Constant
2.12
6.09
Nourinejad, Wenneman, Habib, & Roorda 9
3.2 Parking Simulation Model 1
A PM peak hour parking simulation model is developed for the study area in the Paramics traffic 2
simulation software (26) which models vehicle movements at a microscopic level. The PM peak hour 3
was selected based on field observations showing that this is when the greatest degree of parking 4
activity was occurring. The Toronto CBD experiences greater levels of activity in the PM peak hour 5
because: A large number of workers are commuting out of the CBD at this time; a large number of 6
people are entering the city to shop eat or go to entertainment locations; and trucks in the peak of their 7
deliveries (trucks often avoid the AM peak hour because of congestion). The major inputs to this model 8
are a detailed road network, parking facility locations and capacities, and truck and passenger vehicle 9
demand matrices. 10
The Paramics road network for the study area was extracted from a larger network developed 11
and calibrated for a previous project (27). Parking facility locations were identified in a comprehensive 12
inventory taken in the summer of 2010 (24), and were coded into the simulation network. 13
The data for the development of truck and passenger vehicle demand matrices were retrieved 14
from Toronto’s household travel survey (the Transportation Tomorrow Survey - TTS), City of Toronto 15
intersection traffic counts, and the truck parking survey by Kwok (24). TTS data were used to calculate 16
the passenger vehicle trip generation and attraction for the study area. Truck trip generation and 17
attraction was determined from the truck parking survey. The entry and exit points of inbound and 18
outbound trucks and passenger vehicles were distributed among the roads entering the study area using 19
intersection count data obtained from City of Toronto. Trips through the study area were calculated 20
from the residual intersection counts after inbound and outbound trips had been subtracted. The model 21
assumes no trips had both an origin and destination within the study area. 22
The parking choice model is integrated within the simulation model. The choice model is called 23
each time a vehicle arrives at a potential parking facility which is within 250 m (820 ft) of its final 24
destination. This distance threshold is evident from the parking surveys which show that no freight 25
vehicle was parked further than 250 m away from its destination. The model then calculates the 26
probability of selecting the targeted parking facility. Using a Monte Carlo simulation and the calculated 27
binary choice probability, the vehicle decides whether to park in the facility or to keep driving to find a 28
better parking opportunity. Once parked, vehicles dwell at the facility until they reach their parking 29
duration time when they leave the facility and drive to their next destination outside the study area 30
boundaries. The dwell time for each vehicle is calculated using Monte Carlo simulation and is obtained 31
from a cumulative percentage distribution function of dwell time which is calculated based on a curve 32
fit to the observed data. This function is equal to !!=0.183×exp 6.045×!, where x is a random 33
variable between 0 and 1. Setting x to 1 makes f(x) equal 77 minutes which is the maximum dwell time 34
in the observed data. 35
Figure 2 is a schematic of the simulation process applied simultaneously to each vehicle. The 36
flowchart is interpreted in the following steps: 37
38
1. The simulation model initiates at time T0. 39
2. Vehicles are traced if within 250 m of their final destination. 40
3. Traced vehicles evaluate each parking facility they approach using the binary logit model, until 41
one is chosen. 42
4. When a parking facility is chosen, its capacity is reduced by 1 spot which is taken by the 43
vehicle. Similarly, the capacity of the facility is increased by 1 when the vehicles reaches its 44
dwell time and leaves the facility. 45
5. The model stops tracing vehicles at the time they reach their dwell time and are dispatched from 46
the parking facility to leave the network. 47
6. The model terminates when time reaches the simulation duration which is set to 1.5 hours in 48
this study with 0.5 hours of warm-up. 49
Nourinejad, Wenneman, Habib, & Roorda 10
1
FIGURE 2 Parking simulation model flowchart. 2
3
Three measures of effectiveness calculated in the model are average search time, average 4
walking distance, and total network travel time. Vehicle search time is defined as the difference 5
between the time a vehicle crosses a radius of 250 m of its destination to the time the vehicle finds a 6
spot. Walking distance is defined as the distance between the final destination of the delivery and the 7
Nourinejad, Wenneman, Habib, & Roorda 11
parking location. Intuitively, lower values of both measures of effectiveness are more attractive to both 1
parking authorities and users. Total network travel time (measured in minutes) is the sum of the travel 2
time of all vehicles in the simulation starting from the moment a vehicle enters the study area and 3
ending at the moment it exits. 4
5
4 SCENARIOS AND RESULTS 6
The integrated parking choice-simulation model is designed to evaluate various parking policies. To test 7
the model, we apply the THRU Street parking concept. The two assessed parking policy Scenarios are 8
the following: 9
10
Scenario 1: Sheppard and Temperance Streets are designated as access streets where access to parking 11
facilities is given only to freight vehicles. Richmond and Adelaide Street are designated as THRU 12
streets where freight parking is prohibited (Fig. 3a). 13
14
Scenario 2: Sheppard and Temperance Streets are designated as access streets where access to parking 15
facilities is given only to freight vehicles. Freight vehicles are permitted, however, to park elsewhere in 16
the study area (Fig. 3b). 17
18
FIGURE 3 Scenario 1 (a) and Scenario 2 (b). 19
20
The results of the Scenarios 1 and 2 are compared to the Base Scenario representing the existing 21
parking policy which allows parking of freight and passenger vehicles on all streets of the study area. 22
To account for random variation in the model, 30 runs are executed for each Scenario, and mean and 23
standard deviation of each measure of effectiveness is provided. Table 2 presents the measures of 24
effectiveness for the Base Scenario and two THRU Street Scenarios, for each vehicle type. 25
26
27
28
29
30
31
32
Nourinejad, Wenneman, Habib, & Roorda 12
TABLE 2 Comparison Between Base and THRU Street Scenarios 1
Search Time (minutes)
Walking Distance (metres)
Mean Total
Network
Travel Time
(min)
Scenarios
Freight Vehicles
Passenger Vehicles
Freight Vehicles
Passenger Vehicles
Mean
SD
Mean
SD
Mean
SD
Mean
SD
Base Scenario
2.3
1.27
1.49
1.34
71.5
42.6
84.5
23.2
3542
Scenario 1
1.13*
1.05
1.62
1.41
88.2*
37.2
85.1
19.4
3497
Scenario 2
0.92*
0.98
1.74
1.46
36.8*
32.1
92.3
24.1
3439
Note: changes in means are significantly different from the base Scenario with 95% degree of confidence if an asterisk follows the value
2
Comparison of the three Scenarios shows expected differences between the search time and 3
walking distances of both passenger and commercial vehicles. Scenario 1 results in lower freight search 4
times compared to the Base Scenario, (although the difference is not statistically significant). This is 5
due to the presence of more vacant spots in the access streets that are now available to freight vehicles. 6
The freight vehicle search time standard deviation is also lower for Scenario 1 due to the exclusive 7
access granted to freight vehicles.. In Scenario 2, however, mean freight vehicle search time is further 8
reduced to to 55 seconds, a significant reduction. This happens because those freight vehicles with 9
destinations on THRU streets that were forced to drive to the access streets in Scenario 1 can now drive 10
directly to their final destination. In general, the standard deviation for search times is relatively high, 11
indicating that some vehicles are able to find parking very quickly while some vehicles spend far more 12
time searching for parking. This is consistent with the reality that if a vehicle does not find parking at a 13
close distance the first time they pass their destination, they may spend significant time travelling 14
around the block to make a second attempt. 15
Walking distances for freight vehicles, on the other hand, are higher for Scenario 1 for freight 16
vehicles. This is due to the nature of the policy. Requiring freight vehicles to park on specific access 17
streets restricts the drivers from parking at a location closer to their destination. Hence, drivers have to 18
walk further to reach their final delivery/pickup locations. The mean freight walking distance in 19
Scenario 2, however, is significantly lower. This happens because those vehicles that were restricted in 20
Scenario 1 can now drive to their destinations and park at a closer location. 21
Passenger vehicles, on the other hand, experience different outcomes. Higher mean passenger 22
vehicle searching time results in both Scenarios 1 and 2 (although the differences are not significant). 23
This is due to a diversion of parking demand from the access streets to other locations where parking is 24
harder to find. On the whole, the results of the three Scenarios quantify an expected trade-off between 25
measures of effectiveness of passenger and freight vehicles. 26
Total network travel time can be impacted in three ways. First, the cruising vehicles add to the 27
total travel time. Second, the vehicles that are cruising for parking increase traffic volumes which lead 28
to higher link travel times for all vehicles. Third, vehicles that park on-street decrease the capacity of 29
the link by occupying segments of the rightmost lane. The proposed model is sensitive to the first two 30
but not the third. The last column of Table 2 presents the mean of total network travel time which is 31
lower in Scenarios 1 and 2 compared to the base Scenario. 32
33
5 CONCLUSIONS AND FUTURE DIRECTIONS 34
The integrated parking behaviour-simulation model presented in this paper is a new approach to parking 35
policy evaluation. The model is able to capture important dimensions of parking activity such as 36
walking distance, congestion impact, and parking search times that are commonly neglected in the 37
literature, and usually not quantified at all in practical decision-making. With some effort the method 38
can be applied in any jurisdiction for which a traffic simulation network and appropriate information 39
about parking supply and demand are available. While the most crucial applications are in dense urban 40
Nourinejad, Wenneman, Habib, & Roorda 13
areas where the greatest competition exists for curb space, smaller urban areas with localized parking 1
hotspots are also potential application areas. 2
To verify that the model provides useful and reasonable results, we apply the model to two 3
Scenarios for a small but busy study area in the Toronto CBD. These Scenarios dedicate parking on 4
some interior streets to trucks. Reductions in freight vehicle searching time occur in these Scenarios, 5
whereas freight vehicle walking distances depend on the parking policy for other streets in the network. 6
Passenger vehicle search time and walk distances increase because some of their parking options are 7
removed. All of these changes are intuitive, lending credibility to the model, and they quantitatively 8
illustrate the trade-offs that arise in selecting among competing uses of curb space. 9
The model could be improved and further validated. First, parking spot availability/occupancy 10
driver search time and walking distance were not collected in enough detail for the study area in the 11
parking choice survey. Testing model outcomes against observed values for these critical measures 12
would improve confidence in the model. Second, all trucks are assumed to make parking decisions that 13
conform to a single simple choice function. Couriers, food deliveries and shredding trucks, as 14
examples, all have very different constraints on their parking behaviour that could be represented with 15
more detail if data were available. 16
This research could be further extended to evaluate the effectiveness of other parking policies 17
such as time restrictions, parking information systems, pricing strategies, and new parking facilities, or 18
requirements for new developments. However, some additional data collection efforts may be required 19
for evaluation of these policies. Additional data can be integrated into the simulation by enhancing the 20
parking choice models to include price variables or prior knowledge of parking availability. 21
22
ACKNOWLEDGEMENTS 23
This research was made possible through financial support from the Canadian Transportation Research 24
Forum - Canadian Freight Transportation Research Program. We thank an advisory group that included 25
individuals representing the Canadian Transportation Research Forum, and a local advisory group 26
representing City of Toronto, Metrolinx, BOMA Toronto, and the Canadian Courier and Logistics 27
Association. We gratefully acknowledge the efforts of Justin Kwok, who collected data necessary for 28
this project. 29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
Nourinejad, Wenneman, Habib, & Roorda 14
REFERENCES 1
[1] O'Laughin, R., Rinnan, M. and Thomas, D. (2007, November). Chicago Downtown Freight Study. 2
Transportation Research Board 2008. 3
4
[2] Haider, M., 2009. Challenges Facing Express Delivery Services in Canada’s Urban Centres. 5
6
[3] Federal Highway Administration, 2009. Urban Freight Case Studies. Washington, DC: Federal 7
Highway Administration. 8
9
[4] Habib, K., M., Morency, C., Trepanier, M. 2012. Integrating parking bahaviour in activity-based 10
travel demand modeling: Investigation of the relationship between parking choice and activity 11
scheduling process. Transportation Research Part A (46) 1, 154-166. 12
13
[5] Zalewski, A.J., Buckley, S.M., and Weinberger, R. 2011. Regulating Curb Space: Developing a 14
Framework to Understand and Improve Curbside Management. Proceedings of the 91st Annual 15
Meeting of the Transportation Research Board. Washington DC, January. 16
17
[6] Kolozsvari and Shoup, 2003. Turning small change into big changes. Access (23) 2-7. 18
19
[7] Jaller, M., Holguín-Veras, J., & Hodge, S., 2013. Parking in the City: Challenges for Freight 20
Traffic. Transportation Research Board 92nd Annual Meeting. Washington D.C. 21
22
[8] New York City Department of Transportation., 2004. NY THRU Streets: An Innovative Approach 23
to Managing Midtown Traffic. 24
25
[9] Shoup, D. C., 2006. Cruising for parking. Transport Policy: Special Issue on Parking, 13(6), 47926
486. 27
28
[10] Arnott, R., Inci, E. 2006. An integrated model of downtown parking and traffic congestion. Journal 29
of Urban Economics 60, 418442. 30
31
[11] Gallo, M., D’Acierno, L., Montella, B. 2011. A multiplayer model to simulate cruising for parking 32
in urban areas. Transport Policy (18) 5, 735-744. 33
34
[12] Lam, W., Li, Z-C., Huang, H.-J., Wong, S. 2006. Modeling time-dependent travel choice problems 35
in road networks with multiple user classes and multiple parking facilities. Transportation 36
Research Part B: Methodological, 40, 368395. 37
38
[13] Li, Z., C., Huang, H., j., Lam, W., H., K., Wong, S., C., 2007. A model for evaluation of transport 39
policies in multimodal networks with road and parking capacity constraints. Journal of 40
Mathematical Modelling and Algorithms (6), 239-257. 41
42
[14] Ommeren, J., N., V., Wentink, D., Rietveld, P. 2012. Empirical evidence on cruising for parking. 43
Transportation research part A: Policy and Practice (46) 1, 123-130. 44
45
[15] Benenson I., Martens K., Birfir S., 2008, PARKAGENT: An agent-based model of parking in the 46
city, Computers, Environments and Urban systems, 32, 431-439. 47
48
49
Nourinejad, Wenneman, Habib, & Roorda 15
[16] Dieussaert, K., Aerts, K., Therese, S., Maerivoet, S., Spitaels, K., 2009. SUSTAPARK: An Agent 1
based Model for Simulating Parking Search. 12th AGILE international conference on 2
geographic information science. Leibniz, Germany. 3
4
[17] Hess, S., Train, K., E., Polak, J., W., 2006. On the use of a Modified Latin Hypercube Sampling 5
(MLHS) method in the estimation of a Mixed Logit Model for vehicle choice. Transportation 6
Research Part B: Methodological, 2, 147-163. 7
8
[18] Munuzuri, J., Racero, J., Larraneta, J., 2002. Parking search modeling in freight transport and 9
private traffic simulation. Urban Transport VIII, WIT Press. 10
11
[19] Waraich, R.A. and K.W. Axhausen (2012) An Agent-based Parking Choice Model, paper 12
presented at the 91st Annual Meeting of the Transportation Research Board, Washington, D.C., 13
January 2012. 14
15
[20] Van der Goot, D., 1982. A model to describe the choice of parking places. Transportation 16
Research Part A: General, I(2). 17
18
[21] Axhausen, K.W., Polak, J., 1991. Choice of parking: stated preference approach. Transportation 19
18, 59-81. 20
21
[22] Teknomo, K., & Hokao, K., 1997. Parking Behavior in Central Business District--A Study Case of 22
Surabaya, Indonesia. Easts Journal, 2(2), 551570. 23
24
[23] Thompson, R. G., & Richardson, A. J., 1998. A Parking Search Model. Transportation Research 25
Part A: Policy and Practice, 32(3), 159170. 26
27
[24] Kwok, J. 2010. Data Collection on Parking and Loading Supply and Truck Driver Demand 28
Survey. Final Report. September. 29
30
[25] Ben-Akiva, M., Lerman, S., 1985. Discrete Choice Analysis: Theory and Application to Travel 31
Demand. MIT Press, USA. 32
33
[26] Quadstone Paramics Ltd. Paramics Traffic Microsimulation Software. <http://quadstone-34
paramics.com/> (accessed 05.01.13) 35
36
[27] Amirjamshidi, G, Mostafa, T., S., Misra, A., Roorda, M., J., 2013. Integrated model for 37
microsimulating vehicle emissions, pollutant dispersion and population exposure. 38
Transportation Research Part D: Transport and Environment (18) 16-24. 39
... Freight vehicles have different parking needs compared to passenger vehicles: (a) they need more space since they are larger vehicles and workers need of extra space to access the cargo and unload goods; (b) drivers have a lower threshold for walking since they often carry heavy loads; (c) they have limited access to parking lots if using vehicles with a comparatively higher height and commercial vehicles often cannot use car parks reserved for passenger vehicles; (d) they have shorter parking duration; and (e) they have a limited flexibility in adjusting schedule or travel mode; (f) they are often more willing to park in unauthorized locations (Nourinejad, Wenneman, Habib, & Roorda, 2014), e.g., on street (Demir, Huang, Scholts, & Van Woensel, 2015). Therefore, some urban areas and buildings have been equipped with loading bays for loading/unloading goods. ...
... There are several potentially relevant factors influencing parking choice. Compiled from Axhausen and Polak (1991), Teknomo and Hokao (1997), Waraich and Axhausen (2012), and Nourinejad et al. (2014), examples are: access time, trip purpose, age, gender, search time, queue time, availability, fees, ability to support fine costs, walking time, security and comfort. A model of parking choice for commercial vehicles is estimated by Dalla Chiara et al. (2020) taking into account explanatory variables such as the cost of alternatives (inclusive of expected fines), the total delivery staff, volume to be handled and expected access time. ...
Article
A growing body of research looks specifically at freight vehicle parking choices for purposes of deliveries to street retail, and choice impacts on travel time/uncertainty, congestion, and emissions. However, little attention was given to large urban freight traffic generators, e.g., shopping malls and commercial buildings with offices and retail. These pose different challenges to manage freight vehicle parking demand, due to the limited parking options. To study these, we propose an agent-based simulation approach which integrates data-driven parking-choice models and a demand/supply simulation model. A case study compares demand management strategies (DMS), influencing parking choices, and their impact in reducing freight vehicle parking externalities, such as traffic congestion. DMS include changes to parking capacity, availability, and pricing as well as services (centralized receiving) and technology-based solutions (directed parking). The case study for a commercial region in Singapore shows DMS can improve travel time, parking costs, emission levels and reducing the queuing. This study contributes with a generalizable method, and to local understanding of technology and policy potential. The latter can be of value for managers of large traffic generators and public authorities as a way to understand to select suitable DMS.
... As a direct result of this, many commercial vehicles are forced to park illegally closer to their delivery location, resulting in parking tickets. Parking is easier to find in the off-peak period [6]. ...
Article
Full-text available
Off-peak delivery (OPD) is the delivery of goods during the evening and overnight hours. This strategy has the potential to alleviate peak period congestion, improve efficiency of delivery firms, and reduce emissions. This paper investigates benefits and challenges of a pilot OPD program in the Region of Peel, with the goal of informing potential broader implementations of OPD. In contrast to other previously implemented OPD projects, this OPD pilot focuses on deliveries in suburban areas. Three firms, delivering to 14 pilot retail stores, participated in the OPD pilot in the Region of Peel from March to August 2019. The analysis shows that during the six-month pilot, the average speed of the trips that were made in off-peak hours, from 7:00 p.m. to 7:00 a.m. the next day, is 18.1% faster than those that happened in day-time hours. Furthermore, the total greenhouse gas emissions/km decreased by 10.6%, and emissions factors for air quality pollutants, including CO, NOx, PM10, and PM2.5 reduced by 10.8% to 15.0% in off-peak hours. Results for service times varied between firms, but on average increased by 15.2%, indicating activities in the off-peak hours at the retail stores that prevented overall improvements in service time compared to day-time deliveries. A post-pilot interview was done with logistics managers of the three firms, which provides rich insights about challenges, successes, and ways that the OPD program could be improved.
Article
Full-text available
The choice of freight vehicle type and shipment size are among the most important logistics decisions made by firms. An important aspect is the nature of the choice process i.e., whether the two choices are sequential or joint in nature. In this study, we investigate the factors that influence the two choices and develop sequential and nested logit models with both possible sequence or nesting structures i.e., vehicle type first or at upper level and shipment size second or at lower level and vice-versa. A commercial travel survey for the Greater Toronto and Hamilton Area is used to estimate the models. Characteristics of firms including industry type and employment, and characteristics of shipments including commodity type, destination location, and density value are tested. Shipment size is categorized into four categories and four vehicle types are considered. The results show that both sequences and nesting structures are possible. The nested logit model results show a potential correlation among unobserved components of utility for vehicle types (80%) and shipment sizes (38%) which should be considered. Model performance is assessed using rho-squared and BIC value. The results show that the sequential logit model with shipment size first and vehicle type second sequence has the best model fit. However, based on the strong correlation indicated by the nested logit model for vehicle type nested within shipment size choice (second best model), a model reflecting the joint nature of the choice process might be suitable.
Article
In most dense urban environments in emerging markets, retail deliveries are very fragmented to thousands of nanostores. It is not uncommon for a delivery route to include more than 60 stops. Unloading bays are often blocked by regular traffic. Due to the complex urban environment, it is difficult to estimate the benefits of making unloading bays available. In this study, we conduct a field experiment in an urban field lab of one square kilometer in the downtown of Querétaro, Mexico. During the treatment period of one week, we obtain help from the local traffic police to keep the unloading bays available for unloading only. Using advanced GPS devices and extensive manual field observations, we are able to capture the change in driver behavior and the direct efficiency increases. We find a high efficiency gain, not only in travel time (39%) but also – remarkably – in the total time parked (17%). Corrected for other effects, we estimate a gain of about 44% in total time per delivery. Apart from the insights on unloading benefits, we also provide insights into the method of field experimentation in such a complex environment.
Article
Whilst the widespread adoption of electric vans is necessary to improve urban air quality and reduce carbon emissions, it is also self-evident that adequate charging stations are a precondition. However, the investment case for basic charging stations without public subsidies is challenging. In the context of a London case study, four business models are compared, which integrate solar power generation and new/second-life battery storage system with the basic charging facilities. Considering the uncertainties of electricity tariff and solar generation, the optimal infrastructure investment and operational planning has been formulated as a two-stage stochastic optimization model. The results show that: (i) in the integrated business models, the return on investment and charger installations could be increased by up to 5.39% and 17.06% respectively, and the carbon intensity could be reduced by up to 8.13%; (ii) the nondiscriminatory grant annualized as 50 £ is not sufficient, and a differentiated government subsidy policy may be more conducive to achieving a positive return on investment, such as 50 £ for fast chargers and 100 £ for rapid chargers; (iii) in the integrated business models, fast chargers undertake more vehicle-to-grid electricity exchange with the pattern adoption rate increased by up to 52.38%, while rapid chargers mainly ensure the timely charging completion with the usage frequency increased by up to 2.82%.
Article
Full-text available
Parking is a cumbersome part of auto travel because travelers have to search for a spot and walk from that spot to their final destination. This conventional method of parking will change with the arrival of autonomous vehicles (AV). In the near future, users of AVs get dropped off at their final destination and the occupant-free AVs search for the nearest and most convenient parking spot. Hence, individuals no longer bear the discomfort of cruising for parking while sitting in their vehicle. This paper quantifies the impact of AVs on parking occupancy and traffic flow on a corridor that connects a home zone to a downtown zone. The model considers a heterogeneous group of AVs and conventional vehicles (CV) and captures their parking behavior as they try to minimize their generalized travel costs. Insights are obtained from applying the model to two case studies with uniform and linear parking supply along the corridor. We show that (i) CVs park closer to the downtown zone in order to minimize their walking distance, whereas AVs park farther away from the downtown zone to minimize their parking search time, (ii) AVs experience a lower search time than CVs, and (iii) higher AV penetration rates reduce travel costs for both AVs and CVs.
Article
Full-text available
We propose a workflow for trajectory data mining jointly using well-tested (as opposed to ad hoc) machine learning algorithms and unstructured local knowledge of experts and decision-makers, a common requirement in public agencies and consulting businesses. The key step of the workflow is to condense vehicle trajectory data into an analytics base table (ABT) using a set of features so that general-purpose data mining algorithms can be utilized. The case study extracts context-dependent features from high-frequency truck trajectory data from the State of Texas for analyzing patterns of truck parking in the Statewide highway system and for deriving implications for truck parking regulations and investment decisions. The results show that the approach is suitable for time-efficient implementation and provides valuable inputs for applications related to truck parking studies. This paper does not focus on the deeper understanding of the data in the case study; instead, it focuses on demonstrating how the proposed feature-oriented workflow eases the handling of high-volume trajectory data and improves the trackability of the decision process where data mining algorithms and human expertise interact significantly.
Article
Full-text available
As freight deliveries in cities increase due to retail fragmentation and e-commerce, parking is becoming a more and more relevant part of transportation. In fact, many freight vehicles in cities spend more time parked than they are moving. Moreover, part of the public parking space is shared with passenger vehicles, especially cars. Both arrival processes and parking and delivery processes are stochastic in nature. In order to develop a framework for analysis, we propose a queueing model for an urban parking system consisting of delivery bays and general on-street parking spaces. Freight vehicles may park both in the dedicated bays and in general on-street parking, while passenger vehicles only make use of general on-street parking. Our model allows us to create parsimonious insights into the behavior of a delivery bay parking stretch as part of a limited length of curbside. We are able to find explicit expressions for the relevant performance measures, and formally prove a number of monotonicity results. We further conduct a series of numerical experiments to show more intricate properties that cannot be shown analytically. The model helps us shed light onto the effects of allocating scarce urban curb space to dedicated unloading bays at the expense of general on-street parking. In particular, we show that allocating more space to dedicated delivery bays can also make passenger cars better off.
Article
Dwell time is defined as the time that delivery workers spend performing out-of-vehicle activities while their vehicle is parked. Restricting vehicle dwell time is widely used to manage commercial vehicle parking behavior. However, there is insufficient data to help assess the effectiveness of these restrictions. This makes it difficult for policymakers to account for the complexity of commercial vehicle parking behavior. The current study aims to identify factors correlated with dwell time for commercial vehicles. This is accomplished by using generalized linear models with data collected from five buildings that are known to include commercial vehicle activities in the downtown area of Seattle, Washington, USA. Our models showed that dwell times for buildings with concierge services tended to be shorter. Deliveries of documents also tended to have shorter dwell times than oversized supplies deliveries. Passenger vehicle deliveries had shorter dwell times than deliveries made with vehicles with roll-up doors or swing doors (e.g., vans and trucks). When there were deliveries made to multiple locations within a building, the dwell times were significantly longer than dwell times made to one location in a building. The findings from the presented models demonstrate the potential for improving future parking policies for commercial vehicles by considering data collected from different building types, delivered goods, and vehicle types.
Article
Car parking policies often aim to relieve or better distribute parking demand in central urban areas. Evaluating their effectiveness requires the investigation of traveller behaviour and responses to parking attributes such as availability, cost, walking distance, and time restriction. Parking policies, which are usually cost-intensive to implement, are crucial in solving parking problems and transportation system problems in general. To investigate parking policy properly, developing agent-based parking systems is beneficial. Motivated by a recent parking management system designed for Kingsford, Sydney, this paper introduces an agent-based simulation model to analyse the effects of different parking policies in a realistic dynamic framework. Furthermore, a novel behavioural pricing formulation is integrated into the simulation model, which dynamically seeks to maximise the total utility of all agents in the system with consideration of their travel behaviours. The model is implemented for the Kingsford town centre, to investigate the effects of different parking policies and demand scenarios on parking utilisation and system performance.
Article
Full-text available
Problem, research strategy, and findings: Underpriced and overcrowded curb parking creates problems for everyone except a few lucky drivers who find a cheap space; all the other drivers who cruise to find an open space waste time and fuel, congest traffic, and pollute the air. Overpriced and underoccupied parking also creates problems; when curb spaces remain empty, merchants lose potential customers, workers lose jobs, and cities lose tax revenue. To address these problems, San Francisco has established SFpark, a program that adjusts prices to achieve availability of one or two open spaces per block. To measure how prices affected on-street occupancy, we calculated the price elasticity of demand revealed by over 5,000 price and occupancy changes during the program's first year. Price elasticity has an average value of -0.4, but varies greatly by time of day, location, and several other factors. The average meter price fell 1% during the first year, so SFpark adjusted prices without increasing them overall. This study is the first to use measured occupancy to estimate the elasticity of demand for on-street parking. It also offers the first evaluation of pricing that varies by time of day and location to manage curb parking. Takeaway for practice: San Francisco can improve its program by making drivers more aware of the variable prices, reducing the disabled placard abuse, and introducing seasonal price adjustments. Other cities can incorporate performance parking as a form of congestion pricing.
Article
Full-text available
Parking choice is an essential part of individual transportation; however, many travel demand and traffic simulations do not include parking. This paper reports on a proposal for a simple parking model and describes how this model was implemented into an existing, agent-based traffic simulation. The parking model provides feedback to the traffic simulation so that the overall simulation can react to spatial differences in parking demand and supply. Simulation results of a scenario in the city of Zurich, Switzerland, demonstrated that the model could capture key elements of parking, including capacity and pricing, and could assist with designing parking-focused transport policies. The paper also discusses possible work, such as microsimulation of the search for large-scale parking.
Article
Full-text available
The lack of information about parkers' behavior in choosing a parking location in the Central Business District makes it difficult to develop an effective parking policy. The purpose of this study is to understand parkers' behavior in choosing a parking location in the CBD of Surabaya. Three types of parking location choice models were developed, namely Parking Demand Regression Models, Analytic Hierarchy Process and Multinomial Logit Models. The parkers' behavior in choosing a parking location is mainly influenced by the availability of parking spaces, trip purpose, search & queue time, walking time, parking fee, security, and comfortability.
Article
Full-text available
Traditionally, the parking choice/option is considered to be an important factor in only in the mode choice component of a four-stage travel demand modelling system. However, travel demand modelling has been undergoing a paradigm shift from the traditional trip-based approach to an activity-based approach. The activity-based approach is intended to capture the influences of different policy variables at various stages of activity-travel decision making processes. Parking is a key policy variable that captures land use and transportation interactions in urban areas. It is important that the influences of parking choice on activity scheduling behaviour be identified fully. This paper investigates this issue using a sample data set collected in Montreal, Canada. Parking type choice and activity scheduling decision (start time choice) are modelled jointly in order to identify the effects of parking type choice on activity scheduling behaviour. Empirical investigation gives strong evidence that parking type choice influences activity scheduling process. The empirical findings of this investigation challenge the validity of the traditional conception which considers parking choice as exogenous variable only in the mode choice component of travel demand models.
Article
This paper models traffic at the individual vehicle level, estimates emissions from on-road vehicle sources accounting for drive cycles, estimates how those emissions are dispersed through the atmosphere; and finally estimates the exposed population at times of peak emissions. In the study area, the Toronto Waterfront Area, emissions are highest on the high capacity roadways, and higher in the peak direction of traffic. Pollutant concentrations are higher along the freeways. However, population exposure to these pollutants is highest in the central business district due to the higher population density. Evaluation of scenarios shows significant NOx and HC reduction of 12% and 4% when medium duty diesel trucks are converted to ultra-low emission vehicles.
Article
This paper provides insight into the magnitude of the freight parking problem in large urban areas, and the effectiveness of alternative solution strategies. It does so by estimating the demand for parking using freight trip generation estimates, and the supply of parking on the basis of curb space. The paper discusses freight parking management demand strategies developed by governmental agencies and other organizations. In addition, the paper proposes an approximate methodology to quantify freight parking demand and on-street parking availability. Parking demand is expressed as a function of the freight trip generation of individual establishments, and parking availability is estimated to be a function of curb space dimensions and commercial vehicle characteristics. Empirical findings are provided using New York City as a case study. From the analyses and results, the paper provides a set of policy recommendations to help mitigate the issues identified.
Article
A study has been made of the factors which determine the choice of parking places of visitors to a city center. The starting point of the investigation was the results of a traffic and parking survey held in the central area of Haarlem (The Netherlands) in 1972. To describe the choice of parking places, an attempt was made to specify and test a logit chance model. Possible indicators were: walking time, parking charges, occupation rates of different parking alternatives, possible parking-time restriction, and some “accessibility factors” connected with parking alternatives from which a visitor could make a choice. It became obvious that walking time greatly influenced the visitor's choice. Further, except for the group “loading/unloading”, some preference for “off-street carparks” and “parking garages” could be seen over other categories of parking places. It was also notable that visitors seemed just as inclined to use “illegal parking” as an “ideal” parking place on the street. As for the remaining situations, it seemed that the categories related to the purpose of the trip determined which of the factors played an important part.
Article
In this paper we propose an assignment model on urban networks to simulate parking choices; this model is able to simulate the impact of cruising for parking on traffic congestion. For simulating parking choice and estimating the impact of cruising on road congestion we propose a multi-layer network supply model, where each layer simulates a trip phase (on-car trip between the origin and destination zone, cruising for parking at destination zone and walking egress trip). In this model the cruising time is explicitly simulated on the network. The proposed model is tested on a trial network and on a real-scale network; numerical tests highlighted that the proposed model is able to simulate user parking choice behaviour and the impact of cruising for parking upon road congestion, particularly when the average parking saturation degrees exceed 0.7.